[lojban] Re: New PA-proposal

We have changed the selma'o notation from denoting gumselma'o to denoting pagselma'o. For clarity, this uses an underscore rather than a caret. This both makes the notation more consistent when it changes as well as makes sure that PA always refers to the same words it does now.
Additionally, for clarity, I've suggested a set of lujvo to describe the gumselma'o. Tell me if any of them suck or break any rules.
The hierarchy is now relatively hammered out. It now looks like this:

PA:

PA__: (no, pa, re, ci, vo, mu, xa, ze, bi, so, dau, fei, gai, jau, rei/xei, vai), xo
DUhE: du'e, mo'a, rau, ro, so'a, so'e, so'i, so'o, so'u, no'o [xo'e]
PAI: pai, te'o, tu'o

PIhE

PI: pi, ki'o, ra'e, ji'i
FIhU: fi'u, ka'o
PIhE_: pi'e

CEhI: ce'i, ma'u, me'i, ni'u. za'u, da'a, su'e, su'o, ci’i

One of the funniest things about this is that CEhI is the name of the preceding modifier gumselma'o when it's the least certain one to be in that class. Dealing with that first, I think that whatever we do, we should not define {ce'i} as it is officially -- that is, that {pa no pi re ce'i} is "10.2%". The only reason to make "percent" a succeeding modifier (for which we would need a new pagselma'o) is cultural: we're used to saying "ten percent". This is a terrible reason when PA could have so much order with just three pagselma'o. However, I do have a suggestion if people do not like ce'i as a preceding modifier. We can make it an internal modifier that serves the same function as {pi}, except that it asserts the {pi} position to be two digits to the left. In this way, {pa no ce'i} would still be "10%", but "10.2%" would instead be expressed as {pa no ce'i re}. Note that this version saves a syllable, since {ce'i} would take the place of {pi}. In the event of this conversion I suggest CEhI be renamed to DAhA, since it and {ci'i} are the only words in the class that don't come in symmetric pairs.

Another important discussion point addresses the parsing of PA__ strings. (e.g. {pa no pi re}, i.e. strings composed only of PA__ and PI} Rather than apply a bunch of restrictions to where PI valsi are allowed, (e.g. "{pi} can only be used once in a PA__ string") we say that each PI has to assert a new {pi} position, be it taken from the left string, the right string, or it's own (as in {pi} itself). This is illustrated in the second of three examples at the end of the Google Doc. (On a tangent, how is "second of three examples" supposed to be expressed in Lojban? The simplest thing I can come up with is {lo remoi be lo cimei ku mupli}) The only question left regarding how to parse PA__ strings then becomes whether to use left-grouping or right-grouping to evaluate PI. Right now, everything is as if we use left-grouping. (so that the rightmost {pi} or {ra'e} always "wins") However, I've discovered that this can make parsing something like {pa ji'i re pi no xa} awkward, as left-grouping dictates {<pa [pi] ji'i re [pi]> pi <no xa [pi]>}, which comes out to saying only the digit {re} is uncertain. We can either arbitrarily define some confusing order of operations, or simply use right-grouping. As klaku pointed out way-long-time-ago, PA__ are already semantically interpreted with right-grouping, so it wouldn't be too farfetched to apply PI to PA__ with right-grouping. It would, in fact, make a lot of sense. Note that this wouldn't change how the other PIhE or CEhI are evaluated -- they would still parse the same way, left-to-right, without any explicit grouping needed, just waiting for PA to be parsed for input.